Project

General

Profile

1 11 aaronmk
# Database access
2
3 1869 aaronmk
import copy
4 11 aaronmk
import re
5 3238 aaronmk
import time
6 865 aaronmk
import warnings
7 11 aaronmk
8 300 aaronmk
import exc
9 1909 aaronmk
import dicts
10 1893 aaronmk
import iters
11 1960 aaronmk
import lists
12 3241 aaronmk
import profiling
13 1889 aaronmk
from Proxy import Proxy
14 1872 aaronmk
import rand
15 2217 aaronmk
import sql_gen
16 862 aaronmk
import strings
17 131 aaronmk
import util
18 11 aaronmk
19 832 aaronmk
##### Exceptions
20
21 2804 aaronmk
def get_cur_query(cur, input_query=None):
22 2168 aaronmk
    raw_query = None
23
    if hasattr(cur, 'query'): raw_query = cur.query
24
    elif hasattr(cur, '_last_executed'): raw_query = cur._last_executed
25 2170 aaronmk
26
    if raw_query != None: return raw_query
27 2804 aaronmk
    else: return '[input] '+strings.ustr(input_query)
28 14 aaronmk
29 2170 aaronmk
def _add_cursor_info(e, *args, **kw_args):
30
    '''For params, see get_cur_query()'''
31 2771 aaronmk
    exc.add_msg(e, 'query: '+strings.ustr(get_cur_query(*args, **kw_args)))
32 135 aaronmk
33 300 aaronmk
class DbException(exc.ExceptionWithCause):
34 14 aaronmk
    def __init__(self, msg, cause=None, cur=None):
35 2145 aaronmk
        exc.ExceptionWithCause.__init__(self, msg, cause, cause_newline=True)
36 14 aaronmk
        if cur != None: _add_cursor_info(self, cur)
37
38 2143 aaronmk
class ExceptionWithName(DbException):
39
    def __init__(self, name, cause=None):
40 4491 aaronmk
        DbException.__init__(self, 'for name: '
41
            +strings.as_tt(strings.ustr(name)), cause)
42 2143 aaronmk
        self.name = name
43 360 aaronmk
44 3109 aaronmk
class ExceptionWithValue(DbException):
45
    def __init__(self, value, cause=None):
46 4492 aaronmk
        DbException.__init__(self, 'for value: '
47
            +strings.as_tt(strings.urepr(value)), cause)
48 2240 aaronmk
        self.value = value
49
50 2945 aaronmk
class ExceptionWithNameType(DbException):
51
    def __init__(self, type_, name, cause=None):
52 4491 aaronmk
        DbException.__init__(self, 'for type: '+strings.as_tt(strings.ustr(
53
            type_))+'; name: '+strings.as_tt(name), cause)
54 2945 aaronmk
        self.type = type_
55
        self.name = name
56
57 2306 aaronmk
class ConstraintException(DbException):
58 3345 aaronmk
    def __init__(self, name, cond, cols, cause=None):
59
        msg = 'Violated '+strings.as_tt(name)+' constraint'
60 5447 aaronmk
        if cond != None: msg += ' with condition '+strings.as_tt(cond)
61 3345 aaronmk
        if cols != []: msg += ' on columns: '+strings.as_tt(', '.join(cols))
62
        DbException.__init__(self, msg, cause)
63 2306 aaronmk
        self.name = name
64 3345 aaronmk
        self.cond = cond
65 468 aaronmk
        self.cols = cols
66 11 aaronmk
67 2523 aaronmk
class MissingCastException(DbException):
68 4139 aaronmk
    def __init__(self, type_, col=None, cause=None):
69
        msg = 'Missing cast to type '+strings.as_tt(type_)
70
        if col != None: msg += ' on column: '+strings.as_tt(col)
71
        DbException.__init__(self, msg, cause)
72 2523 aaronmk
        self.type = type_
73
        self.col = col
74
75 5576 aaronmk
class EncodingException(ExceptionWithName): pass
76
77 2306 aaronmk
class DuplicateKeyException(ConstraintException): pass
78 13 aaronmk
79 2306 aaronmk
class NullValueException(ConstraintException): pass
80 13 aaronmk
81 3346 aaronmk
class CheckException(ConstraintException): pass
82
83 3109 aaronmk
class InvalidValueException(ExceptionWithValue): pass
84 2239 aaronmk
85 7112 aaronmk
class InvalidTypeException(ExceptionWithNameType): pass
86
87 2945 aaronmk
class DuplicateException(ExceptionWithNameType): pass
88 2143 aaronmk
89 3419 aaronmk
class DoesNotExistException(ExceptionWithNameType): pass
90
91 89 aaronmk
class EmptyRowException(DbException): pass
92
93 865 aaronmk
##### Warnings
94
95
class DbWarning(UserWarning): pass
96
97 1930 aaronmk
##### Result retrieval
98
99
def col_names(cur): return (col[0] for col in cur.description)
100
101
def rows(cur): return iter(lambda: cur.fetchone(), None)
102
103
def consume_rows(cur):
104
    '''Used to fetch all rows so result will be cached'''
105
    iters.consume_iter(rows(cur))
106
107
def next_row(cur): return rows(cur).next()
108
109
def row(cur):
110
    row_ = next_row(cur)
111
    consume_rows(cur)
112
    return row_
113
114
def next_value(cur): return next_row(cur)[0]
115
116
def value(cur): return row(cur)[0]
117
118
def values(cur): return iters.func_iter(lambda: next_value(cur))
119
120
def value_or_none(cur):
121
    try: return value(cur)
122
    except StopIteration: return None
123
124 2762 aaronmk
##### Escaping
125 2101 aaronmk
126 2573 aaronmk
def esc_name_by_module(module, name):
127
    if module == 'psycopg2' or module == None: quote = '"'
128 2101 aaronmk
    elif module == 'MySQLdb': quote = '`'
129
    else: raise NotImplementedError("Can't escape name for "+module+' database')
130 2500 aaronmk
    return sql_gen.esc_name(name, quote)
131 2101 aaronmk
132
def esc_name_by_engine(engine, name, **kw_args):
133
    return esc_name_by_module(db_engines[engine][0], name, **kw_args)
134
135
def esc_name(db, name, **kw_args):
136
    return esc_name_by_module(util.root_module(db.db), name, **kw_args)
137
138
def qual_name(db, schema, table):
139
    def esc_name_(name): return esc_name(db, name)
140
    table = esc_name_(table)
141
    if schema != None: return esc_name_(schema)+'.'+table
142
    else: return table
143
144 1869 aaronmk
##### Database connections
145 1849 aaronmk
146 2097 aaronmk
db_config_names = ['engine', 'host', 'user', 'password', 'database', 'schemas']
147 1926 aaronmk
148 1869 aaronmk
db_engines = {
149
    'MySQL': ('MySQLdb', {'password': 'passwd', 'database': 'db'}),
150
    'PostgreSQL': ('psycopg2', {}),
151
}
152
153
DatabaseErrors_set = set([DbException])
154
DatabaseErrors = tuple(DatabaseErrors_set)
155
156
def _add_module(module):
157
    DatabaseErrors_set.add(module.DatabaseError)
158
    global DatabaseErrors
159
    DatabaseErrors = tuple(DatabaseErrors_set)
160
161
def db_config_str(db_config):
162
    return db_config['engine']+' database '+db_config['database']
163
164 2448 aaronmk
log_debug_none = lambda msg, level=2: None
165 1901 aaronmk
166 1849 aaronmk
class DbConn:
167 2923 aaronmk
    def __init__(self, db_config, autocommit=True, caching=True,
168 3183 aaronmk
        log_debug=log_debug_none, debug_temp=False, src=None):
169 2915 aaronmk
        '''
170
        @param debug_temp Whether temporary objects should instead be permanent.
171
            This assists in debugging the internal objects used by the program.
172 3183 aaronmk
        @param src In autocommit mode, will be included in a comment in every
173
            query, to help identify the data source in pg_stat_activity.
174 2915 aaronmk
        '''
175 1869 aaronmk
        self.db_config = db_config
176 2190 aaronmk
        self.autocommit = autocommit
177
        self.caching = caching
178 1901 aaronmk
        self.log_debug = log_debug
179 2193 aaronmk
        self.debug = log_debug != log_debug_none
180 2915 aaronmk
        self.debug_temp = debug_temp
181 3183 aaronmk
        self.src = src
182 3074 aaronmk
        self.autoanalyze = False
183 3269 aaronmk
        self.autoexplain = False
184
        self.profile_row_ct = None
185 1869 aaronmk
186 3124 aaronmk
        self._savepoint = 0
187 3120 aaronmk
        self._reset()
188 1869 aaronmk
189
    def __getattr__(self, name):
190
        if name == '__dict__': raise Exception('getting __dict__')
191
        if name == 'db': return self._db()
192
        else: raise AttributeError()
193
194
    def __getstate__(self):
195
        state = copy.copy(self.__dict__) # shallow copy
196 1915 aaronmk
        state['log_debug'] = None # don't pickle the debug callback
197 1869 aaronmk
        state['_DbConn__db'] = None # don't pickle the connection
198
        return state
199
200 3118 aaronmk
    def clear_cache(self): self.query_results = {}
201
202 3120 aaronmk
    def _reset(self):
203 3118 aaronmk
        self.clear_cache()
204 3124 aaronmk
        assert self._savepoint == 0
205 3118 aaronmk
        self._notices_seen = set()
206
        self.__db = None
207
208 2165 aaronmk
    def connected(self): return self.__db != None
209
210 3116 aaronmk
    def close(self):
211 3119 aaronmk
        if not self.connected(): return
212
213 3135 aaronmk
        # Record that the automatic transaction is now closed
214 3136 aaronmk
        self._savepoint -= 1
215 3135 aaronmk
216 3119 aaronmk
        self.db.close()
217 3120 aaronmk
        self._reset()
218 3116 aaronmk
219 3125 aaronmk
    def reconnect(self):
220
        # Do not do this in test mode as it would roll back everything
221
        if self.autocommit: self.close()
222
        # Connection will be reopened automatically on first query
223
224 1869 aaronmk
    def _db(self):
225
        if self.__db == None:
226
            # Process db_config
227
            db_config = self.db_config.copy() # don't modify input!
228 2097 aaronmk
            schemas = db_config.pop('schemas', None)
229 1869 aaronmk
            module_name, mappings = db_engines[db_config.pop('engine')]
230
            module = __import__(module_name)
231
            _add_module(module)
232
            for orig, new in mappings.iteritems():
233
                try: util.rename_key(db_config, orig, new)
234
                except KeyError: pass
235
236
            # Connect
237
            self.__db = module.connect(**db_config)
238
239 3161 aaronmk
            # Record that a transaction is already open
240
            self._savepoint += 1
241
242 1869 aaronmk
            # Configure connection
243 2906 aaronmk
            if hasattr(self.db, 'set_isolation_level'):
244
                import psycopg2.extensions
245
                self.db.set_isolation_level(
246
                    psycopg2.extensions.ISOLATION_LEVEL_READ_COMMITTED)
247 2101 aaronmk
            if schemas != None:
248 2893 aaronmk
                search_path = [self.esc_name(s) for s in schemas.split(',')]
249
                run_query(self, 'SET search_path TO '+(','.join(search_path)),
250
                    log_level=3)
251 1869 aaronmk
252
        return self.__db
253 1889 aaronmk
254 1891 aaronmk
    class DbCursor(Proxy):
255 1927 aaronmk
        def __init__(self, outer):
256 1891 aaronmk
            Proxy.__init__(self, outer.db.cursor())
257 2191 aaronmk
            self.outer = outer
258 1927 aaronmk
            self.query_results = outer.query_results
259 1894 aaronmk
            self.query_lookup = None
260 1891 aaronmk
            self.result = []
261 1889 aaronmk
262 2802 aaronmk
        def execute(self, query):
263 2764 aaronmk
            self._is_insert = query.startswith('INSERT')
264 2797 aaronmk
            self.query_lookup = query
265 2148 aaronmk
            try:
266 3162 aaronmk
                try: cur = self.inner.execute(query)
267 2802 aaronmk
                finally: self.query = get_cur_query(self.inner, query)
268 1904 aaronmk
            except Exception, e:
269
                self.result = e # cache the exception as the result
270
                self._cache_result()
271
                raise
272 3004 aaronmk
273
            # Always cache certain queries
274 3183 aaronmk
            query = sql_gen.lstrip(query)
275 3004 aaronmk
            if query.startswith('CREATE') or query.startswith('ALTER'):
276 3007 aaronmk
                # structural changes
277 3040 aaronmk
                # Rest of query must be unique in the face of name collisions,
278
                # so don't cache ADD COLUMN unless it has distinguishing comment
279
                if query.find('ADD COLUMN') < 0 or query.endswith('*/'):
280 3007 aaronmk
                    self._cache_result()
281 3004 aaronmk
            elif self.rowcount == 0 and query.startswith('SELECT'): # empty
282 2800 aaronmk
                consume_rows(self) # fetch all rows so result will be cached
283 3004 aaronmk
284 2762 aaronmk
            return cur
285 1894 aaronmk
286 1891 aaronmk
        def fetchone(self):
287
            row = self.inner.fetchone()
288 1899 aaronmk
            if row != None: self.result.append(row)
289
            # otherwise, fetched all rows
290 1904 aaronmk
            else: self._cache_result()
291
            return row
292
293
        def _cache_result(self):
294 2948 aaronmk
            # For inserts that return a result set, don't cache result set since
295
            # inserts are not idempotent. Other non-SELECT queries don't have
296
            # their result set read, so only exceptions will be cached (an
297
            # invalid query will always be invalid).
298 1930 aaronmk
            if self.query_results != None and (not self._is_insert
299 1906 aaronmk
                or isinstance(self.result, Exception)):
300
301 1894 aaronmk
                assert self.query_lookup != None
302 1916 aaronmk
                self.query_results[self.query_lookup] = self.CacheCursor(
303
                    util.dict_subset(dicts.AttrsDictView(self),
304
                    ['query', 'result', 'rowcount', 'description']))
305 1906 aaronmk
306 1916 aaronmk
        class CacheCursor:
307
            def __init__(self, cached_result): self.__dict__ = cached_result
308
309 1927 aaronmk
            def execute(self, *args, **kw_args):
310 1916 aaronmk
                if isinstance(self.result, Exception): raise self.result
311
                # otherwise, result is a rows list
312
                self.iter = iter(self.result)
313
314
            def fetchone(self):
315
                try: return self.iter.next()
316
                except StopIteration: return None
317 1891 aaronmk
318 2212 aaronmk
    def esc_value(self, value):
319 2663 aaronmk
        try: str_ = self.mogrify('%s', [value])
320
        except NotImplementedError, e:
321
            module = util.root_module(self.db)
322
            if module == 'MySQLdb':
323
                import _mysql
324
                str_ = _mysql.escape_string(value)
325
            else: raise e
326 2374 aaronmk
        return strings.to_unicode(str_)
327 2212 aaronmk
328 2347 aaronmk
    def esc_name(self, name): return esc_name(self, name) # calls global func
329
330 2814 aaronmk
    def std_code(self, str_):
331
        '''Standardizes SQL code.
332
        * Ensures that string literals are prefixed by `E`
333
        '''
334
        if str_.startswith("'"): str_ = 'E'+str_
335
        return str_
336
337 2665 aaronmk
    def can_mogrify(self):
338 2663 aaronmk
        module = util.root_module(self.db)
339 2665 aaronmk
        return module == 'psycopg2'
340 2663 aaronmk
341 2665 aaronmk
    def mogrify(self, query, params=None):
342
        if self.can_mogrify(): return self.db.cursor().mogrify(query, params)
343
        else: raise NotImplementedError("Can't mogrify query")
344
345 5579 aaronmk
    def set_encoding(self, encoding):
346
        encoding_str = sql_gen.Literal(encoding)
347
        run_query(self, 'SET NAMES '+encoding_str.to_str(self))
348
349 2671 aaronmk
    def print_notices(self):
350 2725 aaronmk
        if hasattr(self.db, 'notices'):
351
            for msg in self.db.notices:
352
                if msg not in self._notices_seen:
353
                    self._notices_seen.add(msg)
354
                    self.log_debug(msg, level=2)
355 2671 aaronmk
356 2793 aaronmk
    def run_query(self, query, cacheable=False, log_level=2,
357 2464 aaronmk
        debug_msg_ref=None):
358 2445 aaronmk
        '''
359 2464 aaronmk
        @param log_ignore_excs The log_level will be increased by 2 if the query
360
            throws one of these exceptions.
361 2664 aaronmk
        @param debug_msg_ref If specified, the log message will be returned in
362
            this instead of being output. This allows you to filter log messages
363
            depending on the result of the query.
364 2445 aaronmk
        '''
365 2167 aaronmk
        assert query != None
366
367 3183 aaronmk
        if self.autocommit and self.src != None:
368 3206 aaronmk
            query = sql_gen.esc_comment(self.src)+'\t'+query
369 3183 aaronmk
370 2047 aaronmk
        if not self.caching: cacheable = False
371 1903 aaronmk
        used_cache = False
372 2664 aaronmk
373 3242 aaronmk
        if self.debug:
374
            profiler = profiling.ItersProfiler(start_now=True, iter_text='row')
375 1903 aaronmk
        try:
376 1927 aaronmk
            # Get cursor
377
            if cacheable:
378 3238 aaronmk
                try: cur = self.query_results[query]
379 1927 aaronmk
                except KeyError: cur = self.DbCursor(self)
380 3238 aaronmk
                else: used_cache = True
381 1927 aaronmk
            else: cur = self.db.cursor()
382
383
            # Run query
384 3238 aaronmk
            try: cur.execute(query)
385 3162 aaronmk
            except Exception, e:
386
                _add_cursor_info(e, self, query)
387
                raise
388 3238 aaronmk
            else: self.do_autocommit()
389 1903 aaronmk
        finally:
390 3242 aaronmk
            if self.debug:
391 3244 aaronmk
                profiler.stop(self.profile_row_ct)
392 3242 aaronmk
393
                ## Log or return query
394
395 4491 aaronmk
                query = strings.ustr(get_cur_query(cur, query))
396 3281 aaronmk
                # Put the src comment on a separate line in the log file
397
                query = query.replace('\t', '\n', 1)
398 3239 aaronmk
399 3240 aaronmk
                msg = 'DB query: '
400 3239 aaronmk
401 3240 aaronmk
                if used_cache: msg += 'cache hit'
402
                elif cacheable: msg += 'cache miss'
403
                else: msg += 'non-cacheable'
404 3239 aaronmk
405 3241 aaronmk
                msg += ':\n'+profiler.msg()+'\n'+strings.as_code(query, 'SQL')
406 3240 aaronmk
407 3237 aaronmk
                if debug_msg_ref != None: debug_msg_ref[0] = msg
408
                else: self.log_debug(msg, log_level)
409 3245 aaronmk
410
                self.print_notices()
411 1903 aaronmk
412
        return cur
413 1914 aaronmk
414 2797 aaronmk
    def is_cached(self, query): return query in self.query_results
415 2139 aaronmk
416 2907 aaronmk
    def with_autocommit(self, func):
417 2801 aaronmk
        import psycopg2.extensions
418
419
        prev_isolation_level = self.db.isolation_level
420 2907 aaronmk
        self.db.set_isolation_level(
421
            psycopg2.extensions.ISOLATION_LEVEL_AUTOCOMMIT)
422 2683 aaronmk
        try: return func()
423 2801 aaronmk
        finally: self.db.set_isolation_level(prev_isolation_level)
424 2683 aaronmk
425 2139 aaronmk
    def with_savepoint(self, func):
426 3137 aaronmk
        top = self._savepoint == 0
427 2171 aaronmk
        savepoint = 'level_'+str(self._savepoint)
428 3137 aaronmk
429 3272 aaronmk
        if self.debug:
430 3273 aaronmk
            self.log_debug('Begin transaction', level=4)
431 3272 aaronmk
            profiler = profiling.ItersProfiler(start_now=True, iter_text='row')
432
433 3160 aaronmk
        # Must happen before running queries so they don't get autocommitted
434
        self._savepoint += 1
435
436 3137 aaronmk
        if top: query = 'START TRANSACTION ISOLATION LEVEL READ COMMITTED'
437
        else: query = 'SAVEPOINT '+savepoint
438
        self.run_query(query, log_level=4)
439
        try:
440
            return func()
441
            if top: self.run_query('COMMIT', log_level=4)
442 2139 aaronmk
        except:
443 3137 aaronmk
            if top: query = 'ROLLBACK'
444
            else: query = 'ROLLBACK TO SAVEPOINT '+savepoint
445
            self.run_query(query, log_level=4)
446
447 2139 aaronmk
            raise
448 2930 aaronmk
        finally:
449
            # Always release savepoint, because after ROLLBACK TO SAVEPOINT,
450
            # "The savepoint remains valid and can be rolled back to again"
451
            # (http://www.postgresql.org/docs/8.3/static/sql-rollback-to.html).
452 3137 aaronmk
            if not top:
453
                self.run_query('RELEASE SAVEPOINT '+savepoint, log_level=4)
454 2930 aaronmk
455
            self._savepoint -= 1
456
            assert self._savepoint >= 0
457
458 3272 aaronmk
            if self.debug:
459
                profiler.stop(self.profile_row_ct)
460 3273 aaronmk
                self.log_debug('End transaction\n'+profiler.msg(), level=4)
461 3272 aaronmk
462 2930 aaronmk
            self.do_autocommit() # OK to do this after ROLLBACK TO SAVEPOINT
463 2191 aaronmk
464
    def do_autocommit(self):
465
        '''Autocommits if outside savepoint'''
466 3135 aaronmk
        assert self._savepoint >= 1
467
        if self.autocommit and self._savepoint == 1:
468 2924 aaronmk
            self.log_debug('Autocommitting', level=4)
469 2191 aaronmk
            self.db.commit()
470 2643 aaronmk
471 3155 aaronmk
    def col_info(self, col, cacheable=True):
472 6900 aaronmk
        module = util.root_module(self.db)
473 6899 aaronmk
        if module == 'psycopg2':
474 6900 aaronmk
            qual_table = sql_gen.Literal(col.table.to_str(self))
475 6899 aaronmk
            col_name_str = sql_gen.Literal(col.name)
476
            try:
477 6900 aaronmk
                type_, is_array, default, nullable = row(run_query(self, '''\
478 6899 aaronmk
SELECT
479
format_type(COALESCE(NULLIF(typelem, 0), pg_type.oid), -1) AS type
480
, typcategory = 'A' AS type_is_array
481 6903 aaronmk
, pg_get_expr(pg_attrdef.adbin, attrelid, true) AS default
482 6899 aaronmk
, NOT pg_attribute.attnotnull AS nullable
483
FROM pg_attribute
484
LEFT JOIN pg_type ON pg_type.oid = atttypid
485
LEFT JOIN pg_attrdef ON adrelid = attrelid AND adnum = attnum
486
WHERE
487 6901 aaronmk
attrelid = '''+qual_table.to_str(self)+'''::regclass
488
AND attname = '''+col_name_str.to_str(self)+'''
489 6899 aaronmk
'''
490 6904 aaronmk
                    , recover=True, cacheable=cacheable, log_level=4))
491 6905 aaronmk
            except (DoesNotExistException, StopIteration):
492
                raise sql_gen.NoUnderlyingTableException(col)
493 6900 aaronmk
            if is_array: type_ = sql_gen.ArrayType(type_)
494 6899 aaronmk
        else:
495
            table = sql_gen.Table('columns', 'information_schema')
496
            cols = [sql_gen.Col('data_type'), sql_gen.Col('udt_name'),
497
                'column_default', sql_gen.Cast('boolean',
498
                sql_gen.Col('is_nullable'))]
499
500
            conds = [('table_name', col.table.name),
501
                ('column_name', strings.ustr(col.name))]
502
            schema = col.table.schema
503
            if schema != None: conds.append(('table_schema', schema))
504
505
            cur = select(self, table, cols, conds, order_by='table_schema',
506
                limit=1, cacheable=cacheable, log_level=4)
507
            try: type_, extra_type, default, nullable = row(cur)
508
            except StopIteration: raise sql_gen.NoUnderlyingTableException(col)
509
            if type_ == 'USER-DEFINED': type_ = extra_type
510
            elif type_ == 'ARRAY':
511
                type_ = sql_gen.ArrayType(strings.remove_prefix('_', extra_type,
512
                    require=True))
513 2643 aaronmk
514 6899 aaronmk
        if default != None: default = sql_gen.as_Code(default, self)
515 2819 aaronmk
        return sql_gen.TypedCol(col.name, type_, default, nullable)
516 2917 aaronmk
517
    def TempFunction(self, name):
518
        if self.debug_temp: schema = None
519
        else: schema = 'pg_temp'
520
        return sql_gen.Function(name, schema)
521 1849 aaronmk
522 1869 aaronmk
connect = DbConn
523
524 832 aaronmk
##### Recoverable querying
525 15 aaronmk
526 5577 aaronmk
def parse_exception(db, e, recover=False):
527
    msg = strings.ustr(e.args[0])
528
    msg = re.sub(r'^(?:PL/Python: )?ValueError: ', r'', msg)
529
530
    match = re.match(r'^invalid byte sequence for encoding "(.+?)":', msg)
531
    if match:
532
        encoding, = match.groups()
533
        raise EncodingException(encoding, e)
534
535 5763 aaronmk
    def make_DuplicateKeyException(constraint, e):
536 5577 aaronmk
        cols = []
537
        cond = None
538
        if recover: # need auto-rollback to run index_cols()
539
            try:
540
                cols = index_cols(db, constraint)
541
                cond = index_cond(db, constraint)
542
            except NotImplementedError: pass
543 5763 aaronmk
        return DuplicateKeyException(constraint, cond, cols, e)
544 5577 aaronmk
545 5763 aaronmk
    match = re.match(r'^duplicate key value violates unique constraint "(.+?)"',
546
        msg)
547
    if match:
548
        constraint, = match.groups()
549
        raise make_DuplicateKeyException(constraint, e)
550
551 5764 aaronmk
    match = re.match(r'^could not create unique index "(.+?)"\n'
552
        r'DETAIL:  Key .+? is duplicated', msg)
553
    if match:
554
        constraint, = match.groups()
555 5823 aaronmk
        raise DuplicateKeyException(constraint, None, [], e)
556 5764 aaronmk
557 5577 aaronmk
    match = re.match(r'^null value in column "(.+?)" violates not-null'
558
        r' constraint', msg)
559
    if match:
560
        col, = match.groups()
561
        raise NullValueException('NOT NULL', None, [col], e)
562
563
    match = re.match(r'^new row for relation "(.+?)" violates check '
564
        r'constraint "(.+?)"', msg)
565
    if match:
566
        table, constraint = match.groups()
567
        constraint = sql_gen.Col(constraint, table)
568
        cond = None
569
        if recover: # need auto-rollback to run constraint_cond()
570
            try: cond = constraint_cond(db, constraint)
571
            except NotImplementedError: pass
572
        raise CheckException(constraint.to_str(db), cond, [], e)
573
574 5723 aaronmk
    match = re.match(r'^(?:invalid input (?:syntax|value)\b[^:]*'
575 5717 aaronmk
        r'|.+? out of range)(?:: "(.+?)")?', msg)
576 5577 aaronmk
    if match:
577
        value, = match.groups()
578 5717 aaronmk
        value = util.do_ignore_none(strings.to_unicode, value)
579
        raise InvalidValueException(value, e)
580 5577 aaronmk
581
    match = re.match(r'^column "(.+?)" is of type (.+?) but expression '
582
        r'is of type', msg)
583
    if match:
584
        col, type_ = match.groups()
585
        raise MissingCastException(type_, col, e)
586
587
    match = re.match(r'^could not determine polymorphic type because '
588
        r'input has type "unknown"', msg)
589
    if match: raise MissingCastException('text', None, e)
590
591 5724 aaronmk
    match = re.match(r'^.+? types (.+?) and (.+?) cannot be matched', msg)
592
    if match:
593
        type0, type1 = match.groups()
594
        raise MissingCastException(type0, None, e)
595 5577 aaronmk
596 7112 aaronmk
    match = re.match(r'^.*?\brelation "(.+?)" is not a (table)', msg)
597
    if match:
598
        name, type_ = match.groups()
599
        raise InvalidTypeException(type_, name, e)
600
601 9458 aaronmk
    typed_name_re = r'^(\S+) ("?)(.+?)\2(?: of relation ".+?")?'
602 9457 aaronmk
        # regexp must be followed with text for .*? to match properly
603 5577 aaronmk
604 9458 aaronmk
    match = re.match(typed_name_re+r'(?:\(.*?)? already exists', msg)
605 5577 aaronmk
    if match:
606 9458 aaronmk
        type_, quote, name = match.groups()
607 5577 aaronmk
        raise DuplicateException(type_, name, e)
608
609
    match = re.match(r'more than one (\S+) named ""(.+?)""', msg)
610
    if match:
611
        type_, name = match.groups()
612
        raise DuplicateException(type_, name, e)
613
614
    match = re.match(typed_name_re+r' does not exist', msg)
615
    if match:
616 9458 aaronmk
        type_, quote, name = match.groups()
617 5683 aaronmk
        if type_ == 'function':
618 5709 aaronmk
            match = re.match(r'^(.+?)\(.*\)$', name)
619 5710 aaronmk
            if match: # includes params, so is call rather than cast to regproc
620 5708 aaronmk
                function_name, = match.groups()
621 5711 aaronmk
                func = sql_gen.Function(function_name)
622 5712 aaronmk
                if function_exists(db, func) and msg.find('CAST') < 0:
623 5683 aaronmk
                    # not found only because of a missing cast
624 5715 aaronmk
                    type_ = function_param0_type(db, func)
625 6301 aaronmk
                    col = None
626 5792 aaronmk
                    if type_ == 'anyelement': type_ = 'text'
627 6301 aaronmk
                    elif type_ == 'hstore': # cast just the value param
628
                        type_ = 'text'
629
                        col = 'value'
630
                    raise MissingCastException(type_, col, e)
631 5577 aaronmk
        raise DoesNotExistException(type_, name, e)
632
633
    raise # no specific exception raised
634
635 2139 aaronmk
def with_savepoint(db, func): return db.with_savepoint(func)
636 11 aaronmk
637 2791 aaronmk
def run_query(db, query, recover=None, cacheable=False, log_level=2,
638
    log_ignore_excs=None, **kw_args):
639 2794 aaronmk
    '''For params, see DbConn.run_query()'''
640 830 aaronmk
    if recover == None: recover = False
641 2464 aaronmk
    if log_ignore_excs == None: log_ignore_excs = ()
642
    log_ignore_excs = tuple(log_ignore_excs)
643 3236 aaronmk
    debug_msg_ref = [None]
644 830 aaronmk
645 3267 aaronmk
    query = with_explain_comment(db, query)
646 3258 aaronmk
647 2148 aaronmk
    try:
648 2464 aaronmk
        try:
649 2794 aaronmk
            def run(): return db.run_query(query, cacheable, log_level,
650 2793 aaronmk
                debug_msg_ref, **kw_args)
651 2796 aaronmk
            if recover and not db.is_cached(query):
652 2464 aaronmk
                return with_savepoint(db, run)
653
            else: return run() # don't need savepoint if cached
654 5890 aaronmk
        except Exception, e:
655
            # Give failed EXPLAIN approximately the log_level of its query
656
            if query.startswith('EXPLAIN'): log_level -= 1
657
658
            parse_exception(db, e, recover)
659 2464 aaronmk
    except log_ignore_excs:
660
        log_level += 2
661
        raise
662
    finally:
663 3236 aaronmk
        if debug_msg_ref[0] != None: db.log_debug(debug_msg_ref[0], log_level)
664 830 aaronmk
665 832 aaronmk
##### Basic queries
666
667 3256 aaronmk
def is_explainable(query):
668
    # See <http://www.postgresql.org/docs/8.3/static/sql-explain.html#AEN57749>
669 3257 aaronmk
    return re.match(r'^(?:SELECT|INSERT|UPDATE|DELETE|VALUES|EXECUTE|DECLARE)\b'
670
        , query)
671 3256 aaronmk
672 3263 aaronmk
def explain(db, query, **kw_args):
673
    '''
674
    For params, see run_query().
675
    '''
676 3267 aaronmk
    kw_args.setdefault('log_level', 4)
677 3263 aaronmk
678 3750 aaronmk
    return strings.ustr(strings.join_lines(values(run_query(db,
679
        'EXPLAIN '+query, recover=True, cacheable=True, **kw_args))))
680 3256 aaronmk
        # not a higher log_level because it's useful to see what query is being
681
        # run before it's executed, which EXPLAIN effectively provides
682
683 3265 aaronmk
def has_comment(query): return query.endswith('*/')
684
685
def with_explain_comment(db, query, **kw_args):
686 3269 aaronmk
    if db.autoexplain and not has_comment(query) and is_explainable(query):
687 3265 aaronmk
        query += '\n'+sql_gen.esc_comment(' EXPLAIN:\n'
688
            +explain(db, query, **kw_args))
689
    return query
690
691 2153 aaronmk
def next_version(name):
692 2163 aaronmk
    version = 1 # first existing name was version 0
693 2586 aaronmk
    match = re.match(r'^(.*)#(\d+)$', name)
694 2153 aaronmk
    if match:
695 2586 aaronmk
        name, version = match.groups()
696
        version = int(version)+1
697 2932 aaronmk
    return sql_gen.concat(name, '#'+str(version))
698 2153 aaronmk
699 2899 aaronmk
def lock_table(db, table, mode):
700
    table = sql_gen.as_Table(table)
701
    run_query(db, 'LOCK TABLE '+table.to_str(db)+' IN '+mode+' MODE')
702
703 3303 aaronmk
def run_query_into(db, query, into=None, add_pkey_=False, **kw_args):
704 2085 aaronmk
    '''Outputs a query to a temp table.
705
    For params, see run_query().
706
    '''
707 2789 aaronmk
    if into == None: return run_query(db, query, **kw_args)
708 2790 aaronmk
709
    assert isinstance(into, sql_gen.Table)
710
711 2992 aaronmk
    into.is_temp = True
712 3008 aaronmk
    # "temporary tables cannot specify a schema name", so remove schema
713
    into.schema = None
714 2992 aaronmk
715 2790 aaronmk
    kw_args['recover'] = True
716 2945 aaronmk
    kw_args.setdefault('log_ignore_excs', (DuplicateException,))
717 2790 aaronmk
718 2916 aaronmk
    temp = not db.debug_temp # tables are permanent in debug_temp mode
719 2790 aaronmk
720
    # Create table
721
    while True:
722
        create_query = 'CREATE'
723
        if temp: create_query += ' TEMP'
724
        create_query += ' TABLE '+into.to_str(db)+' AS\n'+query
725 2385 aaronmk
726 2790 aaronmk
        try:
727
            cur = run_query(db, create_query, **kw_args)
728
                # CREATE TABLE AS sets rowcount to # rows in query
729
            break
730 2945 aaronmk
        except DuplicateException, e:
731 2790 aaronmk
            into.name = next_version(into.name)
732
            # try again with next version of name
733
734 5967 aaronmk
    if add_pkey_: add_pkey_or_index(db, into, warn=True)
735 3075 aaronmk
736
    # According to the PostgreSQL doc, "The autovacuum daemon cannot access and
737
    # therefore cannot vacuum or analyze temporary tables. [...] if a temporary
738
    # table is going to be used in complex queries, it is wise to run ANALYZE on
739
    # the temporary table after it is populated."
740
    # (http://www.postgresql.org/docs/9.1/static/sql-createtable.html)
741
    # If into is not a temp table, ANALYZE is useful but not required.
742 3073 aaronmk
    analyze(db, into)
743 2790 aaronmk
744
    return cur
745 2085 aaronmk
746 2120 aaronmk
order_by_pkey = object() # tells mk_select() to order by the pkey
747
748 2199 aaronmk
distinct_on_all = object() # tells mk_select() to SELECT DISTINCT ON all columns
749
750 8333 aaronmk
def has_subset_func(db, table):
751
    return sql_gen.is_underlying_table(table) and function_exists(db, table)
752
753 3420 aaronmk
def mk_select(db, tables=None, fields=None, conds=None, distinct_on=[],
754 3494 aaronmk
    limit=None, start=None, order_by=order_by_pkey, default_table=None,
755
    explain=True):
756 1981 aaronmk
    '''
757 2121 aaronmk
    @param tables The single table to select from, or a list of tables to join
758 2280 aaronmk
        together, with tables after the first being sql_gen.Join objects
759 1981 aaronmk
    @param fields Use None to select all fields in the table
760 2377 aaronmk
    @param conds WHERE conditions: [(compare_left_side, compare_right_side),...]
761 2379 aaronmk
        * container can be any iterable type
762 2399 aaronmk
        * compare_left_side: sql_gen.Code|str (for col name)
763
        * compare_right_side: sql_gen.ValueCond|literal value
764 2199 aaronmk
    @param distinct_on The columns to SELECT DISTINCT ON, or distinct_on_all to
765
        use all columns
766 2786 aaronmk
    @return query
767 1981 aaronmk
    '''
768 2315 aaronmk
    # Parse tables param
769 2964 aaronmk
    tables = lists.mk_seq(tables)
770 2141 aaronmk
    tables = list(tables) # don't modify input! (list() copies input)
771 2315 aaronmk
    table0 = sql_gen.as_Table(tables.pop(0)) # first table is separate
772 2121 aaronmk
773 2315 aaronmk
    # Parse other params
774 2376 aaronmk
    if conds == None: conds = []
775 2650 aaronmk
    elif dicts.is_dict(conds): conds = conds.items()
776 2379 aaronmk
    conds = list(conds) # don't modify input! (list() copies input)
777 3129 aaronmk
    assert limit == None or isinstance(limit, (int, long))
778
    assert start == None or isinstance(start, (int, long))
779 5527 aaronmk
    if limit == 0: order_by = None
780 8334 aaronmk
    if order_by in (None, order_by_pkey) and has_subset_func(db, table0):
781
        # can use subset function for fast querying at large OFFSET values
782 8335 aaronmk
        run_query(db, 'SET LOCAL enable_sort TO off')
783 8334 aaronmk
        table0 = sql_gen.FunctionCall(table0, limit_=limit, offset_=start)
784
        if limit != 0: limit = None # done by function
785
        start = None # done by function
786
        order_by = None # done by function
787
    elif order_by is order_by_pkey:
788 5641 aaronmk
        if lists.is_seq(distinct_on) and distinct_on: order_by = distinct_on[0]
789
        elif table0 != None: order_by = table_order_by(db, table0, recover=True)
790
        else: order_by = None
791 865 aaronmk
792 2315 aaronmk
    query = 'SELECT'
793 2056 aaronmk
794 2315 aaronmk
    def parse_col(col): return sql_gen.as_Col(col, default_table).to_str(db)
795 2056 aaronmk
796 2200 aaronmk
    # DISTINCT ON columns
797 2233 aaronmk
    if distinct_on != []:
798 2467 aaronmk
        query += '\nDISTINCT'
799 2254 aaronmk
        if distinct_on is not distinct_on_all:
800 2200 aaronmk
            query += ' ON ('+(', '.join(map(parse_col, distinct_on)))+')'
801
802
    # Columns
803 3185 aaronmk
    if query.find('\n') >= 0: whitespace = '\n'
804
    else: whitespace = ' '
805
    if fields == None: query += whitespace+'*'
806 2765 aaronmk
    else:
807
        assert fields != []
808 3185 aaronmk
        if len(fields) > 1: whitespace = '\n'
809
        query += whitespace+('\n, '.join(map(parse_col, fields)))
810 2200 aaronmk
811
    # Main table
812 3185 aaronmk
    if query.find('\n') >= 0 or len(tables) > 0: whitespace = '\n'
813
    else: whitespace = ' '
814 3420 aaronmk
    if table0 != None: query += whitespace+'FROM '+table0.to_str(db)
815 865 aaronmk
816 2122 aaronmk
    # Add joins
817 2271 aaronmk
    left_table = table0
818 2263 aaronmk
    for join_ in tables:
819
        table = join_.table
820 2238 aaronmk
821 2343 aaronmk
        # Parse special values
822
        if join_.type_ is sql_gen.filter_out: # filter no match
823 2376 aaronmk
            conds.append((sql_gen.Col(table_not_null_col(db, table), table),
824 2853 aaronmk
                sql_gen.CompareCond(None, '~=')))
825 2343 aaronmk
826 2467 aaronmk
        query += '\n'+join_.to_str(db, left_table)
827 2122 aaronmk
828
        left_table = table
829
830 865 aaronmk
    missing = True
831 2376 aaronmk
    if conds != []:
832 2576 aaronmk
        if len(conds) == 1: whitespace = ' '
833
        else: whitespace = '\n'
834 2578 aaronmk
        query += '\n'+sql_gen.combine_conds([sql_gen.ColValueCond(l, r)
835
            .to_str(db) for l, r in conds], 'WHERE')
836 2227 aaronmk
    if order_by != None:
837 5642 aaronmk
        query += '\nORDER BY '+sql_gen.as_Col(order_by).to_str(db)
838 3297 aaronmk
    if limit != None: query += '\nLIMIT '+str(limit)
839 865 aaronmk
    if start != None:
840 2467 aaronmk
        if start != 0: query += '\nOFFSET '+str(start)
841 865 aaronmk
842 3494 aaronmk
    if explain: query = with_explain_comment(db, query)
843 3266 aaronmk
844 2786 aaronmk
    return query
845 11 aaronmk
846 2054 aaronmk
def select(db, *args, **kw_args):
847
    '''For params, see mk_select() and run_query()'''
848
    recover = kw_args.pop('recover', None)
849
    cacheable = kw_args.pop('cacheable', True)
850 2442 aaronmk
    log_level = kw_args.pop('log_level', 2)
851 2054 aaronmk
852 2791 aaronmk
    return run_query(db, mk_select(db, *args, **kw_args), recover, cacheable,
853
        log_level=log_level)
854 2054 aaronmk
855 2788 aaronmk
def mk_insert_select(db, table, cols=None, select_query=None, returning=None,
856 3181 aaronmk
    embeddable=False, ignore=False, src=None):
857 1960 aaronmk
    '''
858
    @param returning str|None An inserted column (such as pkey) to return
859 2070 aaronmk
    @param embeddable Whether the query should be embeddable as a nested SELECT.
860 2073 aaronmk
        Warning: If you set this and cacheable=True when the query is run, the
861
        query will be fully cached, not just if it raises an exception.
862 3009 aaronmk
    @param ignore Whether to ignore duplicate keys.
863 3181 aaronmk
    @param src Will be included in the name of any created function, to help
864
        identify the data source in pg_stat_activity.
865 1960 aaronmk
    '''
866 2754 aaronmk
    table = sql_gen.remove_table_rename(sql_gen.as_Table(table))
867 2318 aaronmk
    if cols == []: cols = None # no cols (all defaults) = unknown col names
868 3010 aaronmk
    if cols != None: cols = [sql_gen.to_name_only_col(c, table) for c in cols]
869 2063 aaronmk
    if select_query == None: select_query = 'DEFAULT VALUES'
870 2327 aaronmk
    if returning != None: returning = sql_gen.as_Col(returning, table)
871 2063 aaronmk
872 2497 aaronmk
    first_line = 'INSERT INTO '+table.to_str(db)
873 2063 aaronmk
874 3009 aaronmk
    def mk_insert(select_query):
875
        query = first_line
876 3014 aaronmk
        if cols != None:
877
            query += '\n('+(', '.join((c.to_str(db) for c in cols)))+')'
878 3009 aaronmk
        query += '\n'+select_query
879
880
        if returning != None:
881
            returning_name_col = sql_gen.to_name_only_col(returning)
882
            query += '\nRETURNING '+returning_name_col.to_str(db)
883
884
        return query
885 2063 aaronmk
886 3489 aaronmk
    return_type = sql_gen.CustomCode('unknown')
887
    if returning != None: return_type = sql_gen.ColType(returning)
888 3017 aaronmk
889 3009 aaronmk
    if ignore:
890 3017 aaronmk
        # Always return something to set the correct rowcount
891
        if returning == None: returning = sql_gen.NamedCol('NULL', None)
892
893 3009 aaronmk
        embeddable = True # must use function
894 3010 aaronmk
895 3450 aaronmk
        if cols == None: row = [sql_gen.Col(sql_gen.all_cols, 'row')]
896
        else: row = [sql_gen.Col(c.name, 'row') for c in cols]
897 3092 aaronmk
898 3484 aaronmk
        query = sql_gen.RowExcIgnore(sql_gen.RowType(table), select_query,
899 3497 aaronmk
            sql_gen.ReturnQuery(mk_insert(sql_gen.Values(row).to_str(db))),
900
            cols)
901 3009 aaronmk
    else: query = mk_insert(select_query)
902
903 2070 aaronmk
    if embeddable:
904
        # Create function
905 2513 aaronmk
        function_name = sql_gen.clean_name(first_line)
906 3181 aaronmk
        if src != None: function_name = src+': '+function_name
907 2189 aaronmk
        while True:
908
            try:
909 3451 aaronmk
                func = db.TempFunction(function_name)
910 3489 aaronmk
                def_ = sql_gen.FunctionDef(func, sql_gen.SetOf(return_type),
911
                    query)
912 2194 aaronmk
913 3443 aaronmk
                run_query(db, def_.to_str(db), recover=True, cacheable=True,
914 2945 aaronmk
                    log_ignore_excs=(DuplicateException,))
915 2189 aaronmk
                break # this version was successful
916 2945 aaronmk
            except DuplicateException, e:
917 2189 aaronmk
                function_name = next_version(function_name)
918
                # try again with next version of name
919 2070 aaronmk
920 2337 aaronmk
        # Return query that uses function
921 3009 aaronmk
        cols = None
922
        if returning != None: cols = [returning]
923 3451 aaronmk
        func_table = sql_gen.NamedTable('f', sql_gen.FunctionCall(func), cols)
924
            # AS clause requires function alias
925 3298 aaronmk
        return mk_select(db, func_table, order_by=None)
926 2070 aaronmk
927 2787 aaronmk
    return query
928 2066 aaronmk
929 3074 aaronmk
def insert_select(db, table, *args, **kw_args):
930 2085 aaronmk
    '''For params, see mk_insert_select() and run_query_into()
931 2386 aaronmk
    @param into sql_gen.Table with suggested name of temp table to put RETURNING
932
        values in
933 2072 aaronmk
    '''
934 3141 aaronmk
    returning = kw_args.get('returning', None)
935
    ignore = kw_args.get('ignore', False)
936
937 2386 aaronmk
    into = kw_args.pop('into', None)
938
    if into != None: kw_args['embeddable'] = True
939 2066 aaronmk
    recover = kw_args.pop('recover', None)
940 3141 aaronmk
    if ignore: recover = True
941 2066 aaronmk
    cacheable = kw_args.pop('cacheable', True)
942 2673 aaronmk
    log_level = kw_args.pop('log_level', 2)
943 2066 aaronmk
944 3141 aaronmk
    rowcount_only = ignore and returning == None # keep NULL rows on server
945
    if rowcount_only: into = sql_gen.Table('rowcount')
946
947 3074 aaronmk
    cur = run_query_into(db, mk_insert_select(db, table, *args, **kw_args),
948
        into, recover=recover, cacheable=cacheable, log_level=log_level)
949 3141 aaronmk
    if rowcount_only: empty_temp(db, into)
950 3074 aaronmk
    autoanalyze(db, table)
951
    return cur
952 2063 aaronmk
953 2738 aaronmk
default = sql_gen.default # tells insert() to use the default value for a column
954 2066 aaronmk
955 2063 aaronmk
def insert(db, table, row, *args, **kw_args):
956 2085 aaronmk
    '''For params, see insert_select()'''
957 5050 aaronmk
    ignore = kw_args.pop('ignore', False)
958 5094 aaronmk
    if ignore: kw_args.setdefault('recover', True)
959 5050 aaronmk
960 1960 aaronmk
    if lists.is_seq(row): cols = None
961
    else:
962
        cols = row.keys()
963
        row = row.values()
964 2738 aaronmk
    row = list(row) # ensure that "== []" works
965 1960 aaronmk
966 2738 aaronmk
    if row == []: query = None
967
    else: query = sql_gen.Values(row).to_str(db)
968 1961 aaronmk
969 5050 aaronmk
    try: return insert_select(db, table, cols, query, *args, **kw_args)
970 5057 aaronmk
    except (DuplicateKeyException, NullValueException):
971 5050 aaronmk
        if not ignore: raise
972 5163 aaronmk
        return None
973 11 aaronmk
974 3152 aaronmk
def mk_update(db, table, changes=None, cond=None, in_place=False,
975 3153 aaronmk
    cacheable_=True):
976 2402 aaronmk
    '''
977
    @param changes [(col, new_value),...]
978
        * container can be any iterable type
979
        * col: sql_gen.Code|str (for col name)
980
        * new_value: sql_gen.Code|literal value
981
    @param cond sql_gen.Code WHERE condition. e.g. use sql_gen.*Cond objects.
982 3056 aaronmk
    @param in_place If set, locks the table and updates rows in place.
983
        This avoids creating dead rows in PostgreSQL.
984
        * cond must be None
985 3153 aaronmk
    @param cacheable_ Whether column structure information used to generate the
986 3152 aaronmk
        query can be cached
987 2402 aaronmk
    @return str query
988
    '''
989 3057 aaronmk
    table = sql_gen.as_Table(table)
990
    changes = [(sql_gen.to_name_only_col(c, table), sql_gen.as_Value(v))
991
        for c, v in changes]
992
993 3056 aaronmk
    if in_place:
994
        assert cond == None
995 3058 aaronmk
996 5398 aaronmk
        def col_type(col):
997
            return sql_gen.canon_type(db.col_info(
998
                sql_gen.with_default_table(c, table), cacheable_).type)
999
        changes = [(c, v, col_type(c)) for c, v in changes]
1000 3065 aaronmk
        query = 'ALTER TABLE '+table.to_str(db)+'\n'
1001 5396 aaronmk
        query += ',\n'.join(('ALTER COLUMN '+c.to_str(db)+' TYPE '+t+'\nUSING '
1002
            +v.to_str(db) for c, v, t in changes))
1003 3058 aaronmk
    else:
1004
        query = 'UPDATE '+table.to_str(db)+'\nSET\n'
1005
        query += ',\n'.join((c.to_str(db)+' = '+v.to_str(db)
1006
            for c, v in changes))
1007
        if cond != None: query += '\nWHERE\n'+cond.to_str(db)
1008 3056 aaronmk
1009 3266 aaronmk
    query = with_explain_comment(db, query)
1010
1011 2402 aaronmk
    return query
1012
1013 3074 aaronmk
def update(db, table, *args, **kw_args):
1014 2402 aaronmk
    '''For params, see mk_update() and run_query()'''
1015
    recover = kw_args.pop('recover', None)
1016 3043 aaronmk
    cacheable = kw_args.pop('cacheable', False)
1017 3030 aaronmk
    log_level = kw_args.pop('log_level', 2)
1018 2402 aaronmk
1019 3074 aaronmk
    cur = run_query(db, mk_update(db, table, *args, **kw_args), recover,
1020
        cacheable, log_level=log_level)
1021
    autoanalyze(db, table)
1022
    return cur
1023 2402 aaronmk
1024 3286 aaronmk
def mk_delete(db, table, cond=None):
1025
    '''
1026
    @param cond sql_gen.Code WHERE condition. e.g. use sql_gen.*Cond objects.
1027
    @return str query
1028
    '''
1029
    query = 'DELETE FROM '+table.to_str(db)
1030
    if cond != None: query += '\nWHERE '+cond.to_str(db)
1031
1032
    query = with_explain_comment(db, query)
1033
1034
    return query
1035
1036
def delete(db, table, *args, **kw_args):
1037
    '''For params, see mk_delete() and run_query()'''
1038
    recover = kw_args.pop('recover', None)
1039 3295 aaronmk
    cacheable = kw_args.pop('cacheable', True)
1040 3286 aaronmk
    log_level = kw_args.pop('log_level', 2)
1041
1042
    cur = run_query(db, mk_delete(db, table, *args, **kw_args), recover,
1043
        cacheable, log_level=log_level)
1044
    autoanalyze(db, table)
1045
    return cur
1046
1047 135 aaronmk
def last_insert_id(db):
1048 1849 aaronmk
    module = util.root_module(db.db)
1049 135 aaronmk
    if module == 'psycopg2': return value(run_query(db, 'SELECT lastval()'))
1050
    elif module == 'MySQLdb': return db.insert_id()
1051
    else: return None
1052 13 aaronmk
1053 3490 aaronmk
def define_func(db, def_):
1054
    func = def_.function
1055
    while True:
1056
        try:
1057
            run_query(db, def_.to_str(db), recover=True, cacheable=True,
1058
                log_ignore_excs=(DuplicateException,))
1059
            break # successful
1060
        except DuplicateException:
1061 3495 aaronmk
            func.name = next_version(func.name)
1062 3490 aaronmk
            # try again with next version of name
1063
1064 2394 aaronmk
def mk_flatten_mapping(db, into, cols, preserve=[], as_items=False):
1065 2383 aaronmk
    '''Creates a mapping from original column names (which may have collisions)
1066 2415 aaronmk
    to names that will be distinct among the columns' tables.
1067 2383 aaronmk
    This is meant to be used for several tables that are being joined together.
1068 2415 aaronmk
    @param cols The columns to combine. Duplicates will be removed.
1069
    @param into The table for the new columns.
1070 2394 aaronmk
    @param preserve [sql_gen.Col...] Columns not to rename. Note that these
1071
        columns will be included in the mapping even if they are not in cols.
1072
        The tables of the provided Col objects will be changed to into, so make
1073
        copies of them if you want to keep the original tables.
1074
    @param as_items Whether to return a list of dict items instead of a dict
1075 2383 aaronmk
    @return dict(orig_col=new_col, ...)
1076
        * orig_col: sql_gen.Col(orig_col_name, orig_table)
1077 2392 aaronmk
        * new_col: sql_gen.Col(orig_col_name, into)
1078
        * All mappings use the into table so its name can easily be
1079 2383 aaronmk
          changed for all columns at once
1080
    '''
1081 2415 aaronmk
    cols = lists.uniqify(cols)
1082
1083 2394 aaronmk
    items = []
1084 2389 aaronmk
    for col in preserve:
1085 2390 aaronmk
        orig_col = copy.copy(col)
1086 2392 aaronmk
        col.table = into
1087 2394 aaronmk
        items.append((orig_col, col))
1088
    preserve = set(preserve)
1089
    for col in cols:
1090 2716 aaronmk
        if col not in preserve:
1091 3750 aaronmk
            items.append((col, sql_gen.Col(strings.ustr(col), into, col.srcs)))
1092 2394 aaronmk
1093
    if not as_items: items = dict(items)
1094
    return items
1095 2383 aaronmk
1096 2393 aaronmk
def flatten(db, into, joins, cols, limit=None, start=None, **kw_args):
1097 2391 aaronmk
    '''For params, see mk_flatten_mapping()
1098
    @return See return value of mk_flatten_mapping()
1099
    '''
1100 2394 aaronmk
    items = mk_flatten_mapping(db, into, cols, as_items=True, **kw_args)
1101
    cols = [sql_gen.NamedCol(new.name, old) for old, new in items]
1102 5523 aaronmk
    run_query_into(db, mk_select(db, joins, cols, limit=limit, start=start),
1103
        into=into, add_pkey_=True)
1104 3708 aaronmk
        # don't cache because the temp table will usually be truncated after use
1105 2394 aaronmk
    return dict(items)
1106 2391 aaronmk
1107 3079 aaronmk
##### Database structure introspection
1108 2414 aaronmk
1109 3079 aaronmk
#### Tables
1110
1111 4555 aaronmk
def tables(db, schema_like='public', table_like='%', exact=False,
1112
    cacheable=True):
1113 3079 aaronmk
    if exact: compare = '='
1114
    else: compare = 'LIKE'
1115
1116
    module = util.root_module(db.db)
1117
    if module == 'psycopg2':
1118
        conds = [('schemaname', sql_gen.CompareCond(schema_like, compare)),
1119
            ('tablename', sql_gen.CompareCond(table_like, compare))]
1120
        return values(select(db, 'pg_tables', ['tablename'], conds,
1121 4555 aaronmk
            order_by='tablename', cacheable=cacheable, log_level=4))
1122 3079 aaronmk
    elif module == 'MySQLdb':
1123
        return values(run_query(db, 'SHOW TABLES LIKE '+db.esc_value(table_like)
1124
            , cacheable=True, log_level=4))
1125
    else: raise NotImplementedError("Can't list tables for "+module+' database')
1126
1127 4556 aaronmk
def table_exists(db, table, cacheable=True):
1128 3079 aaronmk
    table = sql_gen.as_Table(table)
1129 4556 aaronmk
    return list(tables(db, table.schema, table.name, True, cacheable)) != []
1130 3079 aaronmk
1131 2426 aaronmk
def table_row_count(db, table, recover=None):
1132 2786 aaronmk
    return value(run_query(db, mk_select(db, table, [sql_gen.row_count],
1133 3298 aaronmk
        order_by=None), recover=recover, log_level=3))
1134 2426 aaronmk
1135 5337 aaronmk
def table_col_names(db, table, recover=None):
1136 5528 aaronmk
    return list(col_names(select(db, table, limit=0, recover=recover,
1137
        log_level=4)))
1138 2414 aaronmk
1139 5383 aaronmk
def table_cols(db, table, *args, **kw_args):
1140
    return [sql_gen.as_Col(strings.ustr(c), table)
1141
        for c in table_col_names(db, table, *args, **kw_args)]
1142
1143 5521 aaronmk
def table_pkey_index(db, table, recover=None):
1144
    table_str = sql_gen.Literal(table.to_str(db))
1145
    try:
1146
        return sql_gen.Table(value(run_query(db, '''\
1147
SELECT relname
1148
FROM pg_index
1149
JOIN pg_class index ON index.oid = indexrelid
1150
WHERE
1151
indrelid = '''+table_str.to_str(db)+'''::regclass
1152
AND indisprimary
1153
'''
1154
            , recover, cacheable=True, log_level=4)), table.schema)
1155
    except StopIteration: raise DoesNotExistException('primary key', '')
1156
1157 5389 aaronmk
def table_pkey_col(db, table, recover=None):
1158 5061 aaronmk
    table = sql_gen.as_Table(table)
1159
1160 5989 aaronmk
    module = util.root_module(db.db)
1161
    if module == 'psycopg2':
1162
        return sql_gen.Col(index_cols(db, table_pkey_index(db, table,
1163
            recover))[0], table)
1164
    else:
1165
        join_cols = ['table_schema', 'table_name', 'constraint_schema',
1166
            'constraint_name']
1167
        tables = [sql_gen.Table('key_column_usage', 'information_schema'),
1168
            sql_gen.Join(
1169
                sql_gen.Table('table_constraints', 'information_schema'),
1170
                dict(((c, sql_gen.join_same_not_null) for c in join_cols)))]
1171
        cols = [sql_gen.Col('column_name')]
1172
1173
        conds = [('constraint_type', 'PRIMARY KEY'), ('table_name', table.name)]
1174
        schema = table.schema
1175
        if schema != None: conds.append(('table_schema', schema))
1176
        order_by = 'position_in_unique_constraint'
1177
1178
        try: return sql_gen.Col(value(select(db, tables, cols, conds,
1179
            order_by=order_by, limit=1, log_level=4)), table)
1180
        except StopIteration: raise DoesNotExistException('primary key', '')
1181 5389 aaronmk
1182 5990 aaronmk
def table_has_pkey(db, table, recover=None):
1183
    try: table_pkey_col(db, table, recover)
1184
    except DoesNotExistException: return False
1185
    else: return True
1186
1187 5389 aaronmk
def pkey_name(db, table, recover=None):
1188
    '''If no pkey, returns the first column in the table.'''
1189 5392 aaronmk
    return pkey_col(db, table, recover).name
1190 832 aaronmk
1191 5390 aaronmk
def pkey_col(db, table, recover=None):
1192 5391 aaronmk
    '''If no pkey, returns the first column in the table.'''
1193 5392 aaronmk
    try: return table_pkey_col(db, table, recover)
1194 5393 aaronmk
    except DoesNotExistException: return table_cols(db, table, recover)[0]
1195 5128 aaronmk
1196 2559 aaronmk
not_null_col = 'not_null_col'
1197 2340 aaronmk
1198
def table_not_null_col(db, table, recover=None):
1199
    '''Name assumed to be the value of not_null_col. If not found, uses pkey.'''
1200 5337 aaronmk
    if not_null_col in table_col_names(db, table, recover): return not_null_col
1201 5388 aaronmk
    else: return pkey_name(db, table, recover)
1202 2340 aaronmk
1203 3348 aaronmk
def constraint_cond(db, constraint):
1204
    module = util.root_module(db.db)
1205
    if module == 'psycopg2':
1206
        table_str = sql_gen.Literal(constraint.table.to_str(db))
1207
        name_str = sql_gen.Literal(constraint.name)
1208
        return value(run_query(db, '''\
1209
SELECT consrc
1210
FROM pg_constraint
1211
WHERE
1212
conrelid = '''+table_str.to_str(db)+'''::regclass
1213
AND conname = '''+name_str.to_str(db)+'''
1214
'''
1215
            , cacheable=True, log_level=4))
1216 5443 aaronmk
    else: raise NotImplementedError("Can't get constraint condition for "
1217
        +module+' database')
1218 3348 aaronmk
1219 5520 aaronmk
def index_exprs(db, index):
1220 3322 aaronmk
    index = sql_gen.as_Table(index)
1221 1909 aaronmk
    module = util.root_module(db.db)
1222
    if module == 'psycopg2':
1223 3322 aaronmk
        qual_index = sql_gen.Literal(index.to_str(db))
1224 5520 aaronmk
        return list(values(run_query(db, '''\
1225 3322 aaronmk
SELECT pg_get_indexdef(indexrelid, generate_series(1, indnatts), true)
1226
FROM pg_index
1227
WHERE indexrelid = '''+qual_index.to_str(db)+'''::regclass
1228 2782 aaronmk
'''
1229
            , cacheable=True, log_level=4)))
1230 5520 aaronmk
    else: raise NotImplementedError()
1231 853 aaronmk
1232 5520 aaronmk
def index_cols(db, index):
1233
    '''Can also use this for UNIQUE constraints, because a UNIQUE index is
1234
    automatically created. When you don't know whether something is a UNIQUE
1235
    constraint or a UNIQUE index, use this function.'''
1236
    return map(sql_gen.parse_expr_col, index_exprs(db, index))
1237
1238 5445 aaronmk
def index_cond(db, index):
1239
    index = sql_gen.as_Table(index)
1240
    module = util.root_module(db.db)
1241
    if module == 'psycopg2':
1242
        qual_index = sql_gen.Literal(index.to_str(db))
1243
        return value(run_query(db, '''\
1244
SELECT pg_get_expr(indpred, indrelid, true)
1245
FROM pg_index
1246
WHERE indexrelid = '''+qual_index.to_str(db)+'''::regclass
1247
'''
1248
            , cacheable=True, log_level=4))
1249
    else: raise NotImplementedError()
1250
1251 5521 aaronmk
def index_order_by(db, index):
1252
    return sql_gen.CustomCode(', '.join(index_exprs(db, index)))
1253
1254
def table_cluster_on(db, table, recover=None):
1255
    '''
1256
    @return The table's cluster index, or its pkey if none is set
1257
    '''
1258
    table_str = sql_gen.Literal(table.to_str(db))
1259
    try:
1260
        return sql_gen.Table(value(run_query(db, '''\
1261
SELECT relname
1262
FROM pg_index
1263
JOIN pg_class index ON index.oid = indexrelid
1264
WHERE
1265
indrelid = '''+table_str.to_str(db)+'''::regclass
1266
AND indisclustered
1267
'''
1268
            , recover, cacheable=True, log_level=4)), table.schema)
1269
    except StopIteration: return table_pkey_index(db, table, recover)
1270
1271
def table_order_by(db, table, recover=None):
1272 8223 aaronmk
    '''
1273
    @return None if table is view, because table_cluster_on() would return None
1274
    '''
1275 5525 aaronmk
    if table.order_by == None:
1276
        try: table.order_by = index_order_by(db, table_cluster_on(db, table,
1277
            recover))
1278
        except DoesNotExistException: pass
1279
    return table.order_by
1280 5521 aaronmk
1281 7242 aaronmk
#### Views
1282
1283
def view_exists(db, view):
1284
    view_str = sql_gen.Literal(view.to_str(db))
1285
    try:
1286
        return value(run_query(db, '''\
1287
SELECT relkind = 'v'
1288
FROM pg_class
1289
WHERE oid = '''+view_str.to_str(db)+'''::regclass
1290
'''
1291
            , cacheable=True, log_level=4))
1292
    except DoesNotExistException: return False
1293
1294 3079 aaronmk
#### Functions
1295
1296
def function_exists(db, function):
1297 3423 aaronmk
    qual_function = sql_gen.Literal(function.to_str(db))
1298
    try:
1299 3425 aaronmk
        select(db, fields=[sql_gen.Cast('regproc', qual_function)],
1300
            recover=True, cacheable=True, log_level=4)
1301 3423 aaronmk
    except DoesNotExistException: return False
1302 4146 aaronmk
    except DuplicateException: return True # overloaded function
1303 3423 aaronmk
    else: return True
1304 3079 aaronmk
1305 5713 aaronmk
def function_param0_type(db, function):
1306
    qual_function = sql_gen.Literal(function.to_str(db))
1307
    return value(run_query(db, '''\
1308
SELECT proargtypes[0]::regtype
1309
FROM pg_proc
1310
WHERE oid = '''+qual_function.to_str(db)+'''::regproc
1311
'''
1312
        , cacheable=True, log_level=4))
1313
1314 3079 aaronmk
##### Structural changes
1315
1316
#### Columns
1317
1318 5020 aaronmk
def add_col(db, table, col, comment=None, if_not_exists=False, **kw_args):
1319 3079 aaronmk
    '''
1320
    @param col TypedCol Name may be versioned, so be sure to propagate any
1321
        renaming back to any source column for the TypedCol.
1322
    @param comment None|str SQL comment used to distinguish columns of the same
1323
        name from each other when they contain different data, to allow the
1324
        ADD COLUMN query to be cached. If not set, query will not be cached.
1325
    '''
1326
    assert isinstance(col, sql_gen.TypedCol)
1327
1328
    while True:
1329
        str_ = 'ALTER TABLE '+table.to_str(db)+' ADD COLUMN '+col.to_str(db)
1330
        if comment != None: str_ += ' '+sql_gen.esc_comment(comment)
1331
1332
        try:
1333
            run_query(db, str_, recover=True, cacheable=True, **kw_args)
1334
            break
1335
        except DuplicateException:
1336 5020 aaronmk
            if if_not_exists: raise
1337 3079 aaronmk
            col.name = next_version(col.name)
1338
            # try again with next version of name
1339
1340
def add_not_null(db, col):
1341
    table = col.table
1342
    col = sql_gen.to_name_only_col(col)
1343
    run_query(db, 'ALTER TABLE '+table.to_str(db)+' ALTER COLUMN '
1344
        +col.to_str(db)+' SET NOT NULL', cacheable=True, log_level=3)
1345
1346 4443 aaronmk
def drop_not_null(db, col):
1347
    table = col.table
1348
    col = sql_gen.to_name_only_col(col)
1349
    run_query(db, 'ALTER TABLE '+table.to_str(db)+' ALTER COLUMN '
1350
        +col.to_str(db)+' DROP NOT NULL', cacheable=True, log_level=3)
1351
1352 2096 aaronmk
row_num_col = '_row_num'
1353
1354 4997 aaronmk
row_num_col_def = sql_gen.TypedCol('', 'serial', nullable=False,
1355 3079 aaronmk
    constraints='PRIMARY KEY')
1356
1357 4997 aaronmk
def add_row_num(db, table, name=row_num_col):
1358
    '''Adds a row number column to a table. Its definition is in
1359
    row_num_col_def. It will be the primary key.'''
1360
    col_def = copy.copy(row_num_col_def)
1361
    col_def.name = name
1362 5021 aaronmk
    add_col(db, table, col_def, comment='', if_not_exists=True, log_level=3)
1363 3079 aaronmk
1364
#### Indexes
1365
1366
def add_pkey(db, table, cols=None, recover=None):
1367
    '''Adds a primary key.
1368
    @param cols [sql_gen.Col,...] The columns in the primary key.
1369
        Defaults to the first column in the table.
1370
    @pre The table must not already have a primary key.
1371
    '''
1372
    table = sql_gen.as_Table(table)
1373 5388 aaronmk
    if cols == None: cols = [pkey_name(db, table, recover)]
1374 3079 aaronmk
    col_strs = [sql_gen.to_name_only_col(v).to_str(db) for v in cols]
1375
1376
    run_query(db, 'ALTER TABLE '+table.to_str(db)+' ADD PRIMARY KEY ('
1377
        +(', '.join(col_strs))+')', recover=True, cacheable=True, log_level=3,
1378
        log_ignore_excs=(DuplicateException,))
1379
1380 2998 aaronmk
def add_index(db, exprs, table=None, unique=False, ensure_not_null_=True):
1381 2688 aaronmk
    '''Adds an index on column(s) or expression(s) if it doesn't already exist.
1382 3356 aaronmk
    Currently, only function calls and literal values are supported expressions.
1383 2998 aaronmk
    @param ensure_not_null_ If set, translates NULL values to sentinel values.
1384 2847 aaronmk
        This allows indexes to be used for comparisons where NULLs are equal.
1385 2538 aaronmk
    '''
1386 2964 aaronmk
    exprs = lists.mk_seq(exprs)
1387 2538 aaronmk
1388 2688 aaronmk
    # Parse exprs
1389
    old_exprs = exprs[:]
1390
    exprs = []
1391
    cols = []
1392
    for i, expr in enumerate(old_exprs):
1393 2823 aaronmk
        expr = sql_gen.as_Col(expr, table)
1394 2688 aaronmk
1395 2823 aaronmk
        # Handle nullable columns
1396 2998 aaronmk
        if ensure_not_null_:
1397 3164 aaronmk
            try: expr = sql_gen.ensure_not_null(db, expr)
1398 2860 aaronmk
            except KeyError: pass # unknown type, so just create plain index
1399 2823 aaronmk
1400 2688 aaronmk
        # Extract col
1401 3002 aaronmk
        expr = copy.deepcopy(expr) # don't modify input!
1402 3356 aaronmk
        col = expr
1403
        if isinstance(expr, sql_gen.FunctionCall): col = expr.args[0]
1404
        expr = sql_gen.cast_literal(expr)
1405
        if not isinstance(expr, (sql_gen.Expr, sql_gen.Col)):
1406 2688 aaronmk
            expr = sql_gen.Expr(expr)
1407 3356 aaronmk
1408 2688 aaronmk
1409
        # Extract table
1410
        if table == None:
1411
            assert sql_gen.is_table_col(col)
1412
            table = col.table
1413
1414 3356 aaronmk
        if isinstance(col, sql_gen.Col): col.table = None
1415 2688 aaronmk
1416
        exprs.append(expr)
1417
        cols.append(col)
1418 2408 aaronmk
1419 2688 aaronmk
    table = sql_gen.as_Table(table)
1420
1421 3005 aaronmk
    # Add index
1422 3148 aaronmk
    str_ = 'CREATE'
1423
    if unique: str_ += ' UNIQUE'
1424
    str_ += ' INDEX ON '+table.to_str(db)+' ('+(
1425
        ', '.join((v.to_str(db) for v in exprs)))+')'
1426
    run_query(db, str_, recover=True, cacheable=True, log_level=3)
1427 2408 aaronmk
1428 5987 aaronmk
def add_pkey_index(db, table): add_index(db, pkey_col(db, table), table)
1429
1430 5765 aaronmk
def add_pkey_or_index(db, table, cols=None, recover=None, warn=False):
1431
    try: add_pkey(db, table, cols, recover)
1432
    except DuplicateKeyException, e:
1433
        if warn: warnings.warn(UserWarning(exc.str_(e)))
1434 5988 aaronmk
        add_pkey_index(db, table)
1435 5765 aaronmk
1436 3083 aaronmk
already_indexed = object() # tells add_indexes() the pkey has already been added
1437
1438
def add_indexes(db, table, has_pkey=True):
1439
    '''Adds an index on all columns in a table.
1440
    @param has_pkey bool|already_indexed Whether a pkey instead of a regular
1441
        index should be added on the first column.
1442
        * If already_indexed, the pkey is assumed to have already been added
1443
    '''
1444 5337 aaronmk
    cols = table_col_names(db, table)
1445 3083 aaronmk
    if has_pkey:
1446
        if has_pkey is not already_indexed: add_pkey(db, table)
1447
        cols = cols[1:]
1448
    for col in cols: add_index(db, col, table)
1449
1450 3079 aaronmk
#### Tables
1451 2772 aaronmk
1452 3079 aaronmk
### Maintenance
1453 2772 aaronmk
1454 3079 aaronmk
def analyze(db, table):
1455
    table = sql_gen.as_Table(table)
1456
    run_query(db, 'ANALYZE '+table.to_str(db), log_level=3)
1457 2934 aaronmk
1458 3079 aaronmk
def autoanalyze(db, table):
1459
    if db.autoanalyze: analyze(db, table)
1460 2935 aaronmk
1461 3079 aaronmk
def vacuum(db, table):
1462
    table = sql_gen.as_Table(table)
1463
    db.with_autocommit(lambda: run_query(db, 'VACUUM ANALYZE '+table.to_str(db),
1464
        log_level=3))
1465 2086 aaronmk
1466 3079 aaronmk
### Lifecycle
1467
1468 3247 aaronmk
def drop(db, type_, name):
1469
    name = sql_gen.as_Name(name)
1470
    run_query(db, 'DROP '+type_+' IF EXISTS '+name.to_str(db)+' CASCADE')
1471 2889 aaronmk
1472 3247 aaronmk
def drop_table(db, table): drop(db, 'TABLE', table)
1473
1474 3082 aaronmk
def create_table(db, table, cols=[], has_pkey=True, col_indexes=True,
1475
    like=None):
1476 2675 aaronmk
    '''Creates a table.
1477 2681 aaronmk
    @param cols [sql_gen.TypedCol,...] The column names and types
1478
    @param has_pkey If set, the first column becomes the primary key.
1479 2760 aaronmk
    @param col_indexes bool|[ref]
1480
        * If True, indexes will be added on all non-pkey columns.
1481
        * If a list reference, [0] will be set to a function to do this.
1482
          This can be used to delay index creation until the table is populated.
1483 2675 aaronmk
    '''
1484
    table = sql_gen.as_Table(table)
1485
1486 3082 aaronmk
    if like != None:
1487
        cols = [sql_gen.CustomCode('LIKE '+like.to_str(db)+' INCLUDING ALL')
1488
            ]+cols
1489 5525 aaronmk
        table.order_by = like.order_by
1490 2681 aaronmk
    if has_pkey:
1491
        cols[0] = pkey = copy.copy(cols[0]) # don't modify input!
1492 2872 aaronmk
        pkey.constraints = 'PRIMARY KEY'
1493 2681 aaronmk
1494 3085 aaronmk
    temp = table.is_temp and not db.debug_temp
1495
        # temp tables permanent in debug_temp mode
1496 2760 aaronmk
1497 3085 aaronmk
    # Create table
1498 3383 aaronmk
    def create():
1499 3085 aaronmk
        str_ = 'CREATE'
1500
        if temp: str_ += ' TEMP'
1501
        str_ += ' TABLE '+table.to_str(db)+' (\n'
1502
        str_ += '\n, '.join(c.to_str(db) for c in cols)
1503 3126 aaronmk
        str_ += '\n);'
1504 3085 aaronmk
1505 7113 aaronmk
        opts = dict(recover=True, cacheable=True, log_level=2,
1506 3383 aaronmk
            log_ignore_excs=(DuplicateException,))
1507 7113 aaronmk
        try: run_query(db, str_, **opts)
1508
        except InvalidTypeException: # try again as view
1509
            run_query_into(db, mk_select(db, like, limit=0), into=table, **opts)
1510 3383 aaronmk
    if table.is_temp:
1511
        while True:
1512
            try:
1513
                create()
1514
                break
1515
            except DuplicateException:
1516
                table.name = next_version(table.name)
1517
                # try again with next version of name
1518
    else: create()
1519 3085 aaronmk
1520 2760 aaronmk
    # Add indexes
1521 2773 aaronmk
    if has_pkey: has_pkey = already_indexed
1522
    def add_indexes_(): add_indexes(db, table, has_pkey)
1523
    if isinstance(col_indexes, list): col_indexes[0] = add_indexes_ # defer
1524
    elif col_indexes: add_indexes_() # add now
1525 2675 aaronmk
1526 3084 aaronmk
def copy_table_struct(db, src, dest):
1527
    '''Creates a structure-only copy of a table. (Does not copy data.)'''
1528 3085 aaronmk
    create_table(db, dest, has_pkey=False, col_indexes=False, like=src)
1529 3084 aaronmk
1530 5529 aaronmk
def copy_table(db, src, dest):
1531
    '''Creates a copy of a table, including data'''
1532
    copy_table_struct(db, src, dest)
1533
    insert_select(db, dest, None, mk_select(db, src))
1534
1535 3079 aaronmk
### Data
1536 2684 aaronmk
1537 2970 aaronmk
def truncate(db, table, schema='public', **kw_args):
1538
    '''For params, see run_query()'''
1539 2777 aaronmk
    table = sql_gen.as_Table(table, schema)
1540 2970 aaronmk
    return run_query(db, 'TRUNCATE '+table.to_str(db)+' CASCADE', **kw_args)
1541 2732 aaronmk
1542 2965 aaronmk
def empty_temp(db, tables):
1543
    tables = lists.mk_seq(tables)
1544 2971 aaronmk
    for table in tables: truncate(db, table, log_level=3)
1545 2965 aaronmk
1546 1968 aaronmk
def empty_db(db, schema='public', **kw_args):
1547
    '''For kw_args, see tables()'''
1548
    for table in tables(db, schema, **kw_args): truncate(db, table, schema)
1549 3094 aaronmk
1550 7179 aaronmk
def distinct_table(db, table, distinct_on, joins=None):
1551 3094 aaronmk
    '''Creates a copy of a temp table which is distinct on the given columns.
1552 5902 aaronmk
    Adds an index on table's distinct_on columns, to facilitate merge joins.
1553 3097 aaronmk
    @param distinct_on If empty, creates a table with one row. This is useful if
1554
        your distinct_on columns are all literal values.
1555 7179 aaronmk
    @param joins The joins to use when creating the new table
1556 3099 aaronmk
    @return The new table.
1557 3094 aaronmk
    '''
1558 7179 aaronmk
    if joins == None: joins = [table]
1559
1560 3099 aaronmk
    new_table = sql_gen.suffixed_table(table, '_distinct')
1561 3411 aaronmk
    distinct_on = filter(sql_gen.is_table_col, distinct_on)
1562 3094 aaronmk
1563 3099 aaronmk
    copy_table_struct(db, table, new_table)
1564 3097 aaronmk
1565
    limit = None
1566
    if distinct_on == []: limit = 1 # one sample row
1567 5903 aaronmk
    else: add_index(db, distinct_on, table) # for join optimization
1568 3097 aaronmk
1569 7179 aaronmk
    insert_select(db, new_table, None, mk_select(db, joins,
1570
        [sql_gen.Col(sql_gen.all_cols, table)], distinct_on=distinct_on,
1571
        order_by=None, limit=limit))
1572 3099 aaronmk
    analyze(db, new_table)
1573 3094 aaronmk
1574 3099 aaronmk
    return new_table